Skip to content

Actions: zeux/calm

Actions

All workflows

Actions

Loading...
Loading

Showing runs from all workflows
77 workflow runs
77 workflow runs

Filter by Event

Filter by Status

Filter by Branch

Filter by Actor

Metal: Fix Gemma inference
build #27: Commit 946c2b8 pushed by zeux
April 13, 2024 20:21 43s main
April 13, 2024 20:21 43s
Metal: Optimize gf4 matmul kernel for denser ALU
build #26: Commit 9f89412 pushed by zeux
April 13, 2024 18:45 37s main
April 13, 2024 18:45 37s
tools: Remove cudabench
build #25: Commit 57e93ef pushed by zeux
April 13, 2024 04:38 41s main
April 13, 2024 04:38 41s
Update README.md
build #24: Commit 3731786 pushed by zeux
April 13, 2024 03:35 54s main
April 13, 2024 03:35 54s
Metal: Validate thread_group_size against pipeline limits
build #23: Commit e02f9be pushed by zeux
April 13, 2024 03:17 55s main
April 13, 2024 03:17 55s
Metal: Remove existing trace before capturing a new one
build #22: Commit 8cb8892 pushed by zeux
April 13, 2024 01:39 48s main
April 13, 2024 01:39 48s
Metal: Encode all input buffers via one setBuffers command
build #21: Commit 7e60499 pushed by zeux
April 12, 2024 21:20 46s main
April 12, 2024 21:20 46s
Metal: Reduce kernel dispatch overhead
build #20: Commit b5d71d7 pushed by zeux
April 12, 2024 18:06 43s main
April 12, 2024 18:06 43s
Update README.md
build #19: Commit 8572db0 pushed by zeux
April 12, 2024 15:54 58s main
April 12, 2024 15:54 58s
Merge pull request #6 from zeux/metal
build #18: Commit 68069e4 pushed by zeux
April 11, 2024 23:11 45s main
April 11, 2024 23:11 45s
Implement initial Metal port
build #17: Pull request #6 synchronize by zeux
April 11, 2024 22:38 46s metal
April 11, 2024 22:38 46s
Implement QKV bias support
build #16: Commit 5876b95 pushed by zeux
April 11, 2024 22:38 42s metal
April 11, 2024 22:38 42s
Implement initial Metal port
build #15: Pull request #6 opened by zeux
April 11, 2024 21:35 44s metal
April 11, 2024 21:35 44s
Make OpenMP dependency optional on macOS
build #14: Commit d85dfa6 pushed by zeux
April 11, 2024 21:32 47s metal
April 11, 2024 21:32 47s
Remove required dependency on openmp headers
build #13: Commit c2d1001 pushed by zeux
April 11, 2024 21:28 40s metal
April 11, 2024 21:28 40s
Add macOS build step to GHA
build #12: Commit 31379af pushed by zeux
April 11, 2024 21:24 39s metal
April 11, 2024 21:24 39s
Use a slightly more efficient decoding process for matmul gf4
build #11: Commit 71316b1 pushed by zeux
April 11, 2024 21:16 42s metal
April 11, 2024 21:16 42s
Implement gf4 weight support for Metal
build #10: Commit 369496b pushed by zeux
April 11, 2024 21:02 41s metal
April 11, 2024 21:02 41s
Further tweaks to reduce ALU pressure
build #9: Commit 7975a7f pushed by zeux
April 11, 2024 00:20 1m 3s main
April 11, 2024 00:20 1m 3s
Decode gf4 values two at a time for half matmul_warppar
build #8: Commit 457adc8 pushed by zeux
April 11, 2024 00:09 53s main
April 11, 2024 00:09 53s
Update README.md
build #7: Commit 44eba3f pushed by zeux
April 10, 2024 19:33 43s main
April 10, 2024 19:33 43s
Add build.yml for GHA runs
build #6: Commit 7df8f39 pushed by zeux
April 10, 2024 01:05 44s main
April 10, 2024 01:05 44s
Try just installing nvcc
build #5: Commit 89d8ca0 pushed by zeux
April 10, 2024 01:01 57s gha
gha
April 10, 2024 01:01 57s
Try to use correct package names
build #4: Commit 0aa23a9 pushed by zeux
April 10, 2024 00:57 2m 9s gha
gha
April 10, 2024 00:57 2m 9s
Try to use NV repository for faster installation
build #3: Commit 7663e1f pushed by zeux
April 10, 2024 00:54 19s gha
gha
April 10, 2024 00:54 19s